Jump to content

  • Log In with Google      Sign In   
  • Create Account

#ActualIcebone1000

Posted 13 November 2012 - 03:55 PM

Hey,

My influence maps are working, one for the bullets and one for the agent, and I already have the agent going to the most "influenced"(most heat) bullet.
Things seems to be working just as expected, witch is awesome.

But I did a stupid mistake..even aiming to have an empty map for now, I did the influence maps thinking on future map obstacles..So Im actually computing it with BFS from each bullet.
I though that by doing this the influence wouldnt pass trough obstacles, but I was a fool, Im computing the influence based on the distance, so what happens is that the obstacles receive 0 influence, but things trough it keep receiving influences as always.. -_-"

example: 100 - 90 - 80 - 0 - 60 - 50 (0 is the wall)

(thats really how my influences are for now, it starts from 100, and a linear decay of 10*distance, I can set it to anything I want thou, but the formula is always:
current - decayrate*dist, I suck at figuring out smarter formulas ).

So now I know that I have to consider the amount of steps/cells, not the straight distance..So I added another data member on my Node struct that I use only for those searchs stuff..And lt works..

now: 100 - 90 - 0 - 50 - 40
----------------80 - 70 - 60

Id like to know if this is really the approach I should be taking..

-------

Second, its a question about BFS (not much an AI topic itself), theres any way to keep track of the steps without storing it in each node added to the open list, but on the algorithm itself?

I tried simply keeping an int steps, and at each loop iteration(just before adding the adjacent cells) I ++ steps, seemed logical to me, but this didnt work, it made an awkward 'leaning towards right' influence map...perhaps Im doing some shit in my BFS (although they are doing their job well)...

I just want to know if its possible to know at what step the current cell/nodes being added during the BFS is, without storing this step value inside the (previous) cells/nodes.

#3Icebone1000

Posted 13 November 2012 - 03:53 PM

Hey,

My influence maps are working, one for the bullets and one for the agent, and I already have the agent going to the most "influenced"(most heat) bullet.
Things seems to be working just as expected, witch is awesome.

But I did a stupid mistake..even aiming to have an empty map for now, I did the influence maps thinking on future map obstacles..So Im actually computing it with BFS from each bullet.
I though that by doing this the influence wouldnt pass trough obstacles, but I was a fool, Im computing the influence based on the distance, so what happens is that the obstacles receive 0 influence, but things trough it keep receiving influences as always.. -_-"

example: 100 - 90 - 80 - 0 - 60 - 50 (0 is the wall)

(thats really how my influences are for now, it starts from 100, and a linear decay of 10*distance, I can set it to anything I want thou, but the formula is always:
current - decayrate*dist, I suck at figuring out smarter formulas ).

So now I know that I have to consider the amount of steps/cells, not the straight distance..So I added another data member on my Node struct that I use only for those searchs stuff..And lt works..

now: 100 - 90 - 0 - 50 - 40
----------------80 - 70 - 60

So first Id like to know if this is really the approach I should be taking..

-------

Second, its a question about BFS (not much an AI topic itself), theres any way to keep track of the steps without storing it in each node added to the open list, but on the algorithm itself?

I tried simply keeping an int steps, and at each loop iteration(just before adding the adjacent cells) I ++ steps, seemed logical to me, but this didnt work, it made an awkward 'leaning towards right' influence map...perhaps Im doing some shit in my BFS (although they are doing their job well)...

I just want to know if its possible to know at what step the current cell/nodes being added during the BFS is, without storing this step value inside the (previous) cells/nodes.

#2Icebone1000

Posted 13 November 2012 - 03:52 PM

Hey,

My influence maps are working, one for the bullets and one for the agent, and I already have the agent going to the most "influenced"(most heat) bullet.
Things seems to be working just as expected, witch is awesome.

But I did a stupid mistake..even aiming to have an empty map for now, I did the influence maps thinking on future map obstacles..So Im actually computing it with BFS from each bullet.
I though that by doing this the influence wouldnt pass trough obstacles, but I was a fool, Im computing the influence based on the distance, so what happens is that the obstacles receive 0 influence, but things trough it keep receiving influences as always.. -_-"

example: 100 - 90 - 80 - 0 - 60 - 50 (0 is the wall)

(thats really how my influences are for now, it starts from 100, and a linear decay of 10*distance, I can set it to anything I want thou, but the formula is always:
current - decayrate*dist, I suck at figuring out smarter formulas ).

So now I know that I have to consider the amount of steps/cells, not the straight distance..So I added another data member on my Node struct that I use only for those searchs stuff..And lt works..

now: 100 - 90 - 0 - 50 - 40
-----------80 - 70 - 60

So first Id like to know if this is really the approach I should be taking..

-------

Second, its a question about BFS (not much an AI topic itself), theres any way to keep track of the steps without storing it in each node added to the open list, but on the algorithm itself?

I tried simply keeping an int steps, and at each loop iteration(just before adding the adjacent cells) I ++ steps, seemed logical to me, but this didnt work, it made an awkward 'leaning towards right' influence map...perhaps Im doing some shit in my BFS (although they are doing their job well)...

I just want to know if its possible to know at what step the current cell/nodes being added during the BFS is, without storing this step value inside the (previous) cells/nodes.

#1Icebone1000

Posted 13 November 2012 - 03:51 PM

Hey,

My influence maps are working, one for the bullets and one for the agent, and I already have the agent going to the most "influenced"(most heat) bullet.
Things seems to be working just as expected, witch is awesome.

But I did a stupid mistake..even aiming to have an empty map for now, I did the influence maps thinking on future map obstacles..So Im actually computing it with BFS from each bullet.
I though that by doing this the influence wouldnt pass trough obstacles, but I was a fool, Im computing the influence based on the distance, so what happens is that the obstacles receive 0 influence, but things trough it keep receiving influences as always.. -_-"

example: 100 - 90 - 80 - 0 - 60 - 50 (0 is the wall)

(thats really how my influences are for now, it starts from 100, and a linear decay of 10*distance, I can set it to anything I want thou, but the formula is always:
current - decayrate*dist, I suck at figuring out smarter formulas ).

So now I know that I have to consider the amount of steps/cells, not the straight distance..So I added another data member on my Node struct that I use only for those searchs stuff..And lt works..

now: 100 - 90 - 0 - 50 - 40
80 - 70 - 60

So first Id like to know if this is really the approach I should be taking..

-------

Second, its a question about BFS (not much an AI topic itself), theres any way to keep track of the steps without storing it in each node added to the open list, but on the algorithm itself?

I tried simply keeping an int steps, and at each loop iteration(just before adding the adjacent cells) I ++ steps, seemed logical to me, but this didnt work, it made an awkward 'leaning towards right' influence map...perhaps Im doing some shit in my BFS (although they are doing their job well)...

I just want to know if its possible to know at what step the current cell/nodes being added during the BFS is, without storing this step value inside the (previous) cells/nodes.

PARTNERS